ARTICLE

https://doi.org/10.1038/s41467-019-11725-5 OPEN Cell morphology and dynamics in dividing radiodurans

Kevin Floc’h1,4, Françoise Lacroix1,4, Pascale Servant2, Yung-Sing Wong 3, Jean-Philippe Kleman 1, Dominique Bourgeois 1 & Joanna Timmins 1

Our knowledge of bacterial originates mostly from studies of rod- or crescent- shaped . Here we reveal that , a relatively large spherical

1234567890():,; bacterium with a multipartite , constitutes a valuable system for the study of the nucleoid in cocci. Using advanced microscopy, we show that D. radiodurans undergoes coordinated morphological changes at both the cellular and nucleoid level as it progresses through its . The nucleoid is highly condensed, but also surprisingly dynamic, adopting multiple configurations and presenting an unusual arrangement in which oriC loci are radially distributed around clustered ter sites maintained at the cell centre. Single-particle tracking and fluorescence recovery after photobleaching studies of the histone-like HU suggest that its loose binding to DNA may contribute to this remarkable plasticity. These findings demonstrate that nucleoid organization is complex and tightly coupled to cell cycle progression in this organism.

1 Univ. Grenoble Alpes, CEA, CNRS, IBS, F-38000 Grenoble, France. 2 Institute for Integrative Biology of the Cell (I2BC), CEA, CNRS, Univ. Paris-Sud, Université Paris-Saclay, 91190 Gif-sur-Yvette, France. 3 Univ. Grenoble Alpes, CNRS, DPM, 38000 Grenoble, France. 4These authors contributed equally: Kevin Floc’h, Françoise Lacroix. Correspondence and requests for materials should be addressed to J.T. (email: [email protected])

NATURE COMMUNICATIONS | (2019) 10:3815 | https://doi.org/10.1038/s41467-019-11725-5 | www.nature.com/naturecommunications 1 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11725-5

n all organisms, genomic DNA is compacted several orders of dynamics of the highly abundant histone-like HU protein, which magnitude and yet must remain accessible for essential DNA- is the major NAP associated with genomic DNA in D. radio- I 31 related processes including DNA replication, repair and durans reveal that it only binds loosely to DNA, which may transcription. In bacteria, packaging of genomic DNA is achieved facilitate the structural rearrangements of the nucleoid. Finally, by several mechanisms, including DNA supercoiling1 and DNA we followed the choreography of the oriC and ter loci of chro- compaction by nucleoid-associated (NAPs), such as mosome 1 during the various stages of the cell cycle and show HU2–4. However, recent studies now indicate that additional that they exhibit very different distributions within the cell with factors, such as molecular crowding and depletion forces, also the oriC loci being radially distributed around the centrally play important roles in determining the volume of the cell located ter sites. Taken together, these findings demonstrate that occupied by the nucleoid5–7, and suggest that cell shape and size the properties of D. radiodurans make it particularly well suited may critically influence nucleoid organization7,8. for the study of nucleoid organization in cocci and provide new, So far, a vast majority of the studies of bacterial nucleoids have compelling evidence, indicating that bacterial nucleoids are focused on three model bacteria, , subtilis complex and dynamic entities that are tightly coupled to cell and Caulobacter crescentus, all of which are either rod- or shape, cell cycle progression and septal growth. crescent-shaped bacteria. Recent developments in single-molecule and genome-wide analytical techniques have started to shed light on how these model bacteria organize, compact and segregate Results their , three processes that are intimately connected. Morphological changes during D. radiodurans cell cycle.To Genetic, microscopy and conformation capture follow the morphological changes occurring during the growth of studies have revealed that nucleoids are organized into micro- D. radiodurans, cells were labelled with the membrane dye Nile and macrodomains9, the origins of which are still unclear Red and deposited on an agarose pad for either time-lapse ima- (reviewed in ref. 10) and have been shown to adopt an overall ging on a spinning-disk confocal microscope or super-resolution helical arrangement11–13, which is distributed along the long- imaging using Point Accumulation for Imaging in Nanoscale itudinal axis filling a large fraction of the cell volume ( ~ 70% in E. Topography (PAINT) microscopy (see Methods and Supple- coli). Mapping of specific chromosome loci such as the origin mentary Fig. 1 for a comparison of the two imaging systems). All (oriC) or the termination sites (ter) have also revealed that cell measurements presented below were extracted from super- chromosomes in model bacteria can be seen to adopt two ste- resolution images (Fig. 1a and Supplementary Fig. 2), whereas the reotypical configurations. Under certain conditions, they may order and duration of the various stages of the cell cycle were arrange longitudinally, along an ori-ter axis with the oriC and ter derived from spinning-disk microscopy images and videos. regions located at opposite poles of the cell, or transversally in Exponentially growing D. radiodurans cells cultured in rich which both the oriC and ter are localized at mid-cell, and the left medium form diads32 that transiently form tetrads before the and right arms are located on either sides2,13–15. start of a new cell cycle (Fig. 1b). D. radiodurans cells divide Although in all bacteria irrespective of their shapes, chromo- sequentially in two perpendicular planes33,34. Starting from an some organization and segregation are tightly coupled to the elliptical and largely symmetric diad (Phase 1), the cells initiate division process, in cocci, the directionality of chromosome seg- their cell cycle by a phase of cell growth (Phase 2) that leads to a regation has additionally been proposed to play a key role in slight invagination at the junctions between the septum 16 positioning the division site . The two most studied cocci are the originating from the previous (S-1) and the cell ovoid Streptococcus pneumoniae and the spherical Staphylococcus periphery. Phase 2 is also characterized by a significant decrease aureus, two major human pathogens. So far, however, the small in the ellipticity of the cells (Fig. 1b and Supplementary Figs. 2 size of these cocci has largely restricted the study of nucleoid and 3). Phase 3 starts with the appearance of bright foci rearrangements and concomitant morphological changes occur- corresponding to the onset of the growth of the new septa (S0). ring during their cell cycle. In contrast, Deinococcus radiodurans These foci appear in the centre of the S-1 septum and in the is a non-pathogenic, relatively large ( ~ 2 µm in diameter) coccus, middle of the opposite peripheral cell wall. We observed that one displaying an outstanding resistance to DNA-damaging agents foci may appear before the other in a given phase 3 cell and septal including , UV light and desiccation17–19. Sev- growth in the two cells forming the diad were frequently seen to eral factors have been proposed to contribute to this outstanding be asynchronous (e.g., cells with asterisk in Fig. 1a). Phase 4 fi phenotype: (i) a highly ef cient and redundant DNA repair corresponds to cells in which the growth of the new S0 septa is machinery, (ii) the presence of numerous partial ( sum of lengths of interior and exterior septa < 80% of P, metabolites18,20–22 that contributes to an increased protection of the distance between the central septum and the opposite side of the proteome, and (iii) the unusual properties of its genome. D. the cell; Supplementary Fig. 2) and in Phase 5 the S0 septa are radiodurans has indeed been shown to possess a complex, mul- almost closed (sum of lengths of interior and exterior septa > tipartite genome composed of four replicons each of which are 80% of P; Supplementary Fig. 2). Finally, in Phase 6 septal growth present in multiple copies ranging from four to ten, depending on is complete and tetrads are formed (Fig. 1b). As the cells progress its growth phase23–25. Moreover, D. radiodurans’ genome has through Phases 3 to 6, the invagination at the diad junction 26 been reported to be more condensed than that of radio-sensitive becomes more pronounced and the length of the S-1 septum bacteria, such as E. coli, and to adopt an unusual ring-like progressively decreases (Supplementary Fig. 3) until disappearing structure27,28. These unusual features have been suggested to completely at the end of Phase 6, as the tetrads fall apart to form facilitate genome maintenance and repair29,30. two diads (Fig. 1b, c). In stationary phase cells, the tetrads no Using both spinning-disk time-lapse microscopy and super- longer separate into diads and the number of Phase 6 cells resolution imaging, we have performed a detailed analysis of the increases markedly (Supplementary Fig. 4). Moreover, a sub- morphological changes that occur at the cellular and nucleoid stantial fraction of these tetrads engages in a new division cycle, level, as D. radiodurans grows and divides in alternating per- but cell cycle progression is very slow and only a small number of pendicular planes. These data reveal that D. radiodurans octads are seen (Supplementary Fig. 4). nucleoids are indeed highly condensed, while remaining sur- Various cell parameters were extracted from our images prisingly dynamic, adopting multiple distinct configurations as (Supplementary Figs. 2 and 3) and were used to determine the the bacterium progresses through its cell cycle. Studies of the average shape, size and volume of cells at each phase of the cell

2 NATURE COMMUNICATIONS | (2019) 10:3815 | https://doi.org/10.1038/s41467-019-11725-5 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11725-5 ARTICLE

abc 4 2 3 5 5

4 4 2 4 m 3 1 μ 2 4 1 1 2 Exponential phase 1.70 4 4 1 4 4 3 4 6 5

4 3 1.73 μm 6 4 3 5 6

de5.0 2

25′50″ 4.0 19% )

3 1 22% ′ ″ ′ ″ 3 m 3.0 15 41 34 49 μ 12% 26% 13% 24% 2.0 9% 11′44″ Cell volume ( Cell volume 6 9% 9% 23% 1.0 13′36″ ns 10% 32′42″ 0.0 5 24% 123456 4 Phases

Fig. 1 Morphological changes occurring during the growth of D. radiodurans. a PAINT imaging of live D. radiodurans cells stained with the membrane dye Nile Red. Numbers inside each cell correspond to their growth phase shown schematically in b. Diads marked with an asterisk are composed of two asynchronized cells. Scale bar: 2 µm. b Schematic representation of the different phases of the D. radiodurans cell cycle in exponentially growing bacteria. c On-scale representation of the average cell sizes and morphologies at the different phases of the cell cycle coloured as in b. Phase 6 cells are very similar to Phase 1 cells. d Changes in cell volume during the cell cycle (N = 185 cells, N > 25 for each phase). Data are represented as mean ± SD. Individual values are shown as dots. Cell volumes were calculated by measuring the cell parameters on PAINT images of exponentially growing wild-type D. radiodurans stained with Nile Red. The mean cell volume progressively increases from a volume of 1.87 ± 0.23 μm3 in Phase 1 to a volume of 3.53 ± 0.41 μm3 in Phase 5 just before cytokinesis. (***P < 0.001, ****P < 0.0001, ns: not significant; statistical test: non-parametric Mann–Whitney). e Duration of each growth phase in exponentially growing D. radiodurans cells. Interior annulus: distribution of exponentially growing cells in the different phases of the cell cycle when observed at a single time point (N = 653). Exterior annulus: duration of phases measured on exponentially growing cells observed by time-lapse imaging (N = 44). Data were collected from two independent experiments. Source data are provided as a Source Data file cycle (Fig. 1c, d). D. radiodurans cells were found to grow linearly obtained for the two approaches and the total cell cycle duration throughout their cell cycle (Supplementary Fig. 5), as seen by the measured in our time-lapse experiments (134 ± 11 min) was close progressive increase in cell volume starting from a mean volume to the doubling time of 130 min derived from our growth curves of 1.87 ± 0.23 μm3 in Phase 1 until a volume of 3.53 ± 0.41 μm3 in performed on D. radiodurans cells grown in liquid medium in a Phase 5 just before cytokinesis (Fig. 1d). The change in cell shaking incubator (Supplementary Fig. 6). Cells spend 91% of volume was also accompanied by a change in ellipticity of the their time as diads and 9% as tetrads, and septal growth (Phases 3, cells as they progress through the cell cycle (Supplementary 4 and 5) occurs during more than half of the cell cycle. Phases 1, 5 Fig. 3). Starting from hemispheres in Phase 1, individual cells and 6 are the shortest, lasting between 11 and 15 min, whereas become quasi ellipsoids by the end of Phase 5. In contrast, in Phases 2, 3 and 4 are longer and last 25 to 35 min (Fig. 1e). stationary phase, cell growth arrests and the mean volume of cells in tetrads decreases to 1.62 ± 0.29 μm3 (Supplementary Fig. 4). Cell growth occurs in both peripheral and septal cell walls.In To assess the duration of each growth phase in exponential bacteria, cell growth is associated with expansion of the cell wall cells, two approaches were used. Durations were either deduced and peptidoglycan (PG) synthesis and remodelling. We thus from the distribution of phases in the population of exponentially investigated whether growth of the cell wall occurs (i) in the growing cells when deposited on an agarose pad and directly septal regions only, (ii) throughout the septal and peripheral cell observed under the microscope (the fraction of cells in each phase walls, or (iii) only in the peripheral cell walls of D. radiodurans reflects the relative duration of these phases; Fig. 1e interior (Fig. 2a). For this, cells were pulse labelled with two dyes: annulus) or by time-lapse imaging of cells (Fig. 1e exterior BODIPY-FL 3-amino-D-alanine (BADA), a green dye that is annulus and Supplementary Movie 1). Very similar results were stably incorporated into the pentapeptide chain during PG

NATURE COMMUNICATIONS | (2019) 10:3815 | https://doi.org/10.1038/s41467-019-11725-5 | www.nature.com/naturecommunications 3 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11725-5

a 2nd cycle T = 130 min

i

1st cycle T = 0 Septa only

All cell wall ii

Peripheral cell wall only

iii

b d 6

4 m) μ

c Length ( 2

0 T = 0 min T ≈ 130 min T = 0 min T ≈ 130 min Green labelled cell wall Red labelled cell wall

Fig. 2 Cell growth occurs in both peripheral and septal cell walls. a Schematic representation of three possible models of cell wall growth for a cell initially in Phase 1 and undergoing a complete cell cycle. At T = 0 min, only the peripheral cell wall of the mother cell is stably labelled with BADA (green); the whole cell wall (peripheral and septal) is labelled with Nile Red (red). Nile Red signal suffers from a poorer signal/noise ratio than BADA; thus, the superposition of both colours does not appear yellow. Considering only the periphery of the new diad after a complete cell cycle (T = 130 min), if cell wall growth occurred (i) in the septal regions only, the length of the BADA-labelled cell wall would remain the same, whereas the length of the cell wall labelled only with Nile Red would increase; (ii) throughout the bacterial cell walls, both the length of the BADA-labelled cell wall and the cell wall labelled only with Nile Red would increase; (iii) only in peripheral cell walls, the length of the BADA-labelled cell wall would increase, whereas the length of the cell wall labelled only with Nile Red would remain the same. At T = 130 min, measurements were made on the peripheral cell walls only, which were originally labelled at T = 0. The newly grown septal regions at T = 130 min ( in a, but stained with Nile Red in c) were not taken into account in these analyses. b, c Examples of D. radiodurans cells at T = 0(b) and T = 130 min (c). Cells marked with an asterisk are examples of cells in Phase 1 that were used for the perimeter measurements and cell wall growth analyses (d). Scale bar: 2 µm. d Mean lengths of the green- and red-only-labelled cell walls at T = 0 min (N = 53) and T = 130 min (N = 40) for Phase 1 cells. Both significantly increase after a complete cell cycle. Data are represented as mean ± SD. Individual values are shown as dots. (****P < 0.0001; statistical test: non-parametric Mann–Whitney). Data were collected from three independent experiments. Source data are provided as a Source Data file

4 NATURE COMMUNICATIONS | (2019) 10:3815 | https://doi.org/10.1038/s41467-019-11725-5 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11725-5 ARTICLE

Table 1 Description of bacterial strains used in this study

Strains Description Source or reference E. coli BL21 (DE3) fhuA2 lon ompT gal (λDE3) dcm ΔhsdS Laboratory stock D. radiodurans GY9613 ATCC 13939, R1 Laboratory stock D. radiodurans GY15743 hbs::cherryΩcat Laboratory stock 25 D. radiodurans GY15787 hbs::cherryΩtet, locus ori::kan-parSc2Bc/pFAP246 (Pspac-parBc2Bc::drGFP, cat) 25 D. radiodurans GY15800 hbs::cherryΩtet, locus ter::kan-parSc2Bc/pFAP246 (Pspac-parBc2Bc::drGFP, cat) D. radiodurans GY17031 hbs::PAmcherryΩkan 54 synthesis in the peripheral cell walls only and does not reorganize labelling (Fig. 3 and Supplementary Fig. 8) and all four replicons during the cell cycle, and Nile Red that diffuses into lipid of D. radiodurans assembled into a single nucleoid entity. When membranes and thus labels the whole cell membrane, including using Syto9, cell membranes could additionally be stained with in newly synthesized septa during the cell cycle. The samples were Nile Red, to distinguish the different phases of the cell cycle then washed to remove any unbound dye prior to imaging. The (Fig. 3). use of two lasers prevented D. radiodurans from growing under Nucleoid conformations were assessed in Syto9 and Nile Red- the microscope, so cells were observed shortly after the labelling stained wild-type D. radiodurans cells and representative shapes procedure (T = 0) and then again after a complete cell cycle (T = of nucleoids were retrieved for each phase of the cell cycle (Fig. 3). 130 min). The mode of cell wall growth could be deduced from In addition, time-lapse 3D imaging of either Syto9-stained such images by comparing the lengths of green- and red-only- nucleoids (Supplementary Movie 2) or HU-mCherry-expressing labelled cell walls per cell (Fig. 2a) at T = 0 with that observed D. radiodurans (Supplementary Movie 3) was performed during a after a complete cell cycle (T = 130 min). If growth is restricted to complete cell cycle to determine the chronological order in which the septal region (model i), which is not labelled with the BADA these different conformations appeared. These videos clearly dye at T = 0, the length of the green-labelled cell wall should highlight the dynamic nature of the nucleoids. In Phases 1 and 2, remain constant after one cell cycle; if growth occurs throughout a majority of nucleoids (respectively 52% and 42%) adopted all cell walls (model ii), the lengths of both the green- and red- toroidal shapes with an average diameter of 0.95 ± 0.10 µm labelled cell walls should increase; and if growth is restricted to (Supplementary Fig. 2). In addition, in Phase 1, a substantial the outer periphery of the cells (model iii), then the red-only- fraction (28%) of nucleoids were found to form condensed, labelled cell wall should remain constant (Fig. 2a). For ease of undefined structures, which were seen to be short-lived (minutes) analysis, we performed our measurements on Phase 1 cells that in our time-lapses and rapidly became toroids, which appeared to displayed complete BADA (green) labelling of their peripheral be more stable and long-lasting. Due to the reduced resolution of = cell wall, but only Nile Red staining of their S-1 septum at T 0 the spinning-disk images in the Z-axis, we cannot rule out that (Fig. 2b) and could thus be compared with our proposed models these undefined structures may correspond to top views of toroids presented in Fig. 2a. After a complete cell cycle, Phase 1 cells with an interior ring size below the resolution limit of the images. clearly displayed an altered labelling pattern (Fig. 2b, c). Mea- In Phase 2, 16% of nucleoids presented a square configuration surements revealed that the lengths of both green- and with bright vertices and a significant fraction (21%) also adopted red-labelled regions had significantly increased (Fig. 2d), thus a more open configuration (open ring or crescent). In Phase 3, indicating that cell wall growth occurs throughout the entire cell these two nucleoid configurations were found to be the most surface of D. radiodurans cells including septal regions (Fig. 2a), prevalent (respectively 25% and 30%) and the crescent-shaped 35 as reported for S. aureus . nucleoids were positioned with their convex sides facing the S-1 Next, we evaluated the position and angle of the newly growing septum of the diad (Fig. 3a). The fraction of toroidal-shaped fi S0 septa relative to the previous S-1 septa in exponentially growing nucleoids decreased signi cantly in Phase 3 and instead they were cells. In all cells, septal growth was found to occur precisely in the seen to open up to form crescents. In some cases, an intermediary middle (50% ± 1.5%) at a 90° ± 7° angle to the S-1 septum step was observed in which the toroids became squares with three (Supplementary Fig. 7). Early electron micrographs of D. or four of their vertices exhibiting strong fluorescent foci before radiodurans cells suggested that septal closure occurs through a opening up to adopt a crescent shape. In Phase 4 cells, most of closing door mechanism and not as a diaphragm33. Three- these configurations disappeared and instead a large proportion dimensional (3D) imaging of Nile Red-stained cells confirmed of nucleoids were now elongated rods (54%) with an average this (Supplementary Fig. 7). The closing septum grows from both length of 1.48 ± 0.25 µm and an apparent full width at half sides of the cells leaving a gap that stretches across the whole cell. maximum of 0.52 ± 0.08 µm (Supplementary Fig. 2). More complex branched structures (35%) were also observed in Phase Nucleoid organization during the cell cycle. To follow the 3D 4. Remarkably, these structures were all positioned with their long morphology of D. radiodurans nucleoids as a function of their cell axis perpendicular to the future division axis (Fig. 3a). Time-lapse cycle in exponentially growing and stationary phase bacteria, we imaging revealed that these elongated structures resulted from the used spinning-disk confocal microscopy on live cells. The opening and stretching of the crescent-shaped nucleoids seen in nucleoids were visualized either by using a genetically modified Phase 3. In Phase 5 cells, in which the formation of the newly of D. radiodurans in which the encoding the highly dividing septa was nearly complete, nucleoids appeared mostly as abundant HU protein was endogenously fused to mCherry branched, elongated structures (35%), as seen in Phase 4, or as (Table 1), or by staining the DNA with the Syto9 dye (Fig. 3). The double rings connected by a thin thread (42%), in which each ring high extent of colocalization observed when double staining D. was positioned in one of the two future daughter cells. A clear radiodurans nucleoids with both HU-mCherry and Syto9 (Sup- constriction was observed around the centre of these nucleoids plementary Fig. 8) confirmed that HU largely associates with resulting from the two closing septal segments (Fig. 3a). Time- genomic DNA. Importantly, the same nucleoid structures lapse imaging revealed that the rod-like structures predominantly (described below), including the previously described ring-like or found in Phase 4 progressively formed more complex, branched toroid structure, were observed irrespective of the mode of structures that eventually formed double crescents or rings

NATURE COMMUNICATIONS | (2019) 10:3815 | https://doi.org/10.1038/s41467-019-11725-5 | www.nature.com/naturecommunications 5 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11725-5

ab

Double Toroid Square Crescent Rod Branched rings 1

6 2

5 3

4

c Non defined Toroid Square Crescent Rod Branched Double rings

100%

80%

60%

40%

20% % of each structure per phase

0% 123456 Phases

Fig. 3 Nucleoid organization in exponentially growing D. radiodurans. a Array of the representative shapes of D. radiodurans nucleoids stained with the DNA dye Syto9. Cell membranes were stained with Nile Red. In each box, the representative shape is seen in the upper cell of the diad. In the case of tetrads, the representative shape is the cell marked with an asterisk. Scale: 2 µm. b Snapshot of the on-scale simulated reconstitution of the changes in cell shape and nucleoid structure occurring during the D. radiodurans cell cycle, illustrating the tight coordination of nucleoid rearrangements with septal growth (Supplementary Movie 4). Scale bar: 1 µm. c Distribution of the various nucleoid morphologies observed as a function of the phases of the cell cycle (N = 232 nucleoids; N > 26 for each phase). Source data are provided as a Source Data file connected by short stretches of DNA. Finally, in Phase 6 cells, ( ~ 0.7 μm3) during Phases 1 to 3 and then to increase in Phases 4 chromosome segregation was complete and the newly formed and 5, to reach a maximum volume of ~ 1.2 μm3 just before nucleoids presented similar configurations as those seen in Phase cytokinesis. The newly formed nucleoids in Phase 6 exhibited the 1 cells, i.e., condensed, undefined structures (45%) or toroids smallest volumes ( ~ 0.6 μm3). As a result, the fraction of the cell (45%). These findings are summarized in a simulated movie that occupied by the nucleoid varied as a function of the cell cycle illustrates the principal morphological changes of the nucleoids from a minimal value of 0.30 ± 0.06 observed in Phase 4 to a observed in exponentially growing D. radiodurans cells as they maximal value of 0.36 ± 0.06 detected in Phase 1 cells (Fig. 4b). In progress through their cell cycle (Fig. 3b and Supplementary comparison, the mean cell and nucleoid volumes of E. coli cells Movie 4). Interestingly, as observed previously when characteriz- were found to be 1.33 μm3 and 0.83 μm3, respectively, corre- ing the different phases of the cell cycle, the two cells composing a sponding to a cell fraction occupied by the nucleoid of 0.65 diad were not necessarily synchronized in terms of nucleoid (Supplementary Fig. 7), in agreement with previously reported morphology. However, it was clear that the morphological values11. These quantitative measurements thus confirm that D. changes at the nucleoid level were highly coordinated with cell radiodurans possesses a highly condensed nucleoid occupying cycle progression and septal growth (Fig. 3b). approximately one-third of the cell volume, but also reveal that its level of compaction varies during its cell cycle. We also deter- mined the mean volume of Syto9-stained nucleoids in D. radio- Nucleoid compaction. Three-dimensional images of Syto9- and durans expressing HU-mCherry and found that these were Nile Red-stained D. radiodurans were also used to determine the slightly higher than that of wild-type cells (Supplementary Fig. 8), volume occupied by the nucleoid in D. radiodurans as a function perhaps as a consequence of steric hindrance caused by the of its cell cycle (Fig. 4a; Methods). For comparison, the same mCherry label, which may alter the DNA-binding properties of procedure was used for analysis of E. coli cells (Table 1 and HU. Exponentially growing D. radiodurans cells have been Supplementary Fig. 9). Unlike the cell volume (Fig. 1c), the reported to possess four to ten copies of their 3.2 Mbp genome25, nucleoid volume did not increase linearly throughout the so the DNA density in an average D. radiodurans nucleoid − cell cycle (Fig. 4a). Instead, it was found to remain constant ( ~ 0.7 μm3) should be in the range of 18–46 Mbp µm 3.In

6 NATURE COMMUNICATIONS | (2019) 10:3815 | https://doi.org/10.1038/s41467-019-11725-5 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11725-5 ARTICLE

ab2.0 0.5

) ns 0.4 3

m 1.5 ns μ 0.3 1.0 0.2

0.5 Fraction of cell volume Fraction

Nucleoid volume ( Nucleoid volume 0.1 ns ns 0.0 0.0 123456 123456 Phases Phases

Fig. 4 Nucleoid compaction as a function of D. radiodurans cell cycle. a Changes in nucleoid volume as cells progress through their cell cycle. Nucleoid volumes were retrieved from the three-dimensional signal of Syto9-labelled nucleoids of live, exponentially growing D. radiodurans cells. The nucleoid volume is relatively constant during the first three phases ( ~ 0.7 μm3) and then progressively increases to reach a maximal volume of ~ 1.2 μm3 in Phase 5 cells just before cytokinesis. b The fraction of the cell volume occupied by the nucleoid was retrieved by dividing the nucleoid volume of individual cells by their matched cell volume, derived from spinning-disk images of dual-labelled Syto9- and Nile Red-stained cells, using the cell measurements presented in Supplementary Fig. 2. The fraction of the cell occupied by the nucleoid significantly decreases during the first three phases (from 0.35 to 0.30), and then remains around 0.30 until cytokinesis. After septal closure, the fraction increases again to 0.34. a, b Data are represented as mean ± SD (N = 195, N >29 for each phase). Individual values are shown as dots. (*P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, ns: not significant, statistical test: non-parametric Mann–Whitney). Source data are provided as a Source Data file

E. coli, which possesses one to two copies of a 4.6 Mbp genome, live, exponentially growing D. radiodurans. Single-particle- the DNA density in its nucleoid should be between 5.5 and tracking PALM (sptPALM) experiments were performed on the 11 Mbp µm−3, corresponding to a density at least three times HU-PAmCherry-expressing D. radiodurans strain in which the lower than that observed in D. radiodurans cells. trajectories of individual HU-PAmCherry molecules were Finally, Syto9- and Nile Red-stained stationary phase D. reconstituted (Fig. 5b). Cumulative probability distribution radiodurans tetrads in Phase 6 were also analysed to determine (CPD) analysis37 of 1355 individual tracks extracted from several the level of nucleoid compaction in such cells (Supplementary datasets revealed that only a single population of molecules could Fig. 4). The average fraction of the cell occupied by the nucleoid be distinguished, which displayed a confined diffusion (with a was found to be 0.19, which is significantly lower than the highest confinement radius of ~ 400 nm; Fig. 5b) with a mean apparent level of DNA compaction measured in exponentially growing diffusion coefficient of ~ 0.32 µm2 s−1. This value, which is Phase 4 cells (Fig. 4b). significantly lower than that obtained for freely diffusing fluorescent proteins in bacteria (6-9 µm2 s−1)38 but also higher than that of immobile, DNA-binding proteins (D* Nucleoid and HU dynamics. To further characterize D. radio- < 0.2 µm2 s−1 39,40), clearly indicates that HU is not tightly durans nucleoids in exponentially growing bacteria, we used associated with the genomic DNA. Ensemble measurements of super-resolution photoactivated localization microscopy (PALM) HU mobility probed by fluorescence recovery after photobleach- to image a D. radiodurans strain in which the most abundant ing (FRAP) on HU-mCherry-expressing D. radiodurans cells NAP, HU, was fused to PAmCherry (Table 1). PALM experi- confirmed these findings and revealed that 99% of HU molecules ments were performed exclusively on live bacteria, because fixa- were mobile exhibiting a half- of 1.03 s (Fig. 5c, d). These data tion protocols strongly affected the nucleoid organization (loss of are consistent with a FRAP study of the E. coli NAP, H-NS38, and diverse shapes and decondensation of the DNA). Several of the suggest that even though HU largely colocalizes with genomic nucleoid morphologies described above (e.g., crescent- and rod- DNA, its association with DNA is only transient, which may shaped nucleoids) could be seen in the reconstructed PALM allow the nucleoid to rapidly reorganize as the cells progress images, but most of the distinguishing features and details were through their cell cycle. lost (Fig. 5a). Although PALM provides higher spatial resolution, acquiring a complete dataset in PALM microscopy takes several minutes on our set-up. The loss of details in our PALM images Distribution of oriC and ter sites during the cell cycle.To thus most likely reflects the dynamic nature of the nucleoids in follow the choreography of the oriC and ter loci during the cell the second-to-minute timescale, as confirmed by a spinning-disk cycle of exponentially growing D. radiodurans, we used engi- microscopy time-lapse experiment in which images of HU- neered D. radiodurans strains, GY15787 and GY15800, in which mCherry were acquired in live D. radiodurans with short interval a heterologous ParABS system is used to fluorescently label times (Supplementary Movie 5). This time-lapse video reveals respectively the oriC and ter loci of chromosome 1 (Table 1), that even though most of the nucleoid conformations last for allowing them to be detected as bright foci within the cells25. several minutes (Fig. 3 and Supplementary Movies 2 and 3), the Three-colour 3D spinning-disk images were recorded on such nucleoids are mobile and can be seen to rotate and reorganize in cells, to visualize green fluorescent protein (GFP)-labelled oriC or the second timescale. ter loci, hydroxycoumarin-amino-D-alanine (HADA)-stained cell HU has been shown to play a major role in the compaction and walls and HU-mCherry-labelled nucleoids (Fig. 6a, d). When organization of the nucleoid in D. radiodurans25,31 and to be analysing these images, we chose to restrict our analysis to the essential for viability36. We therefore wondered whether HU may bright ParB-labelled foci that could unambiguously be dis- contribute to the remarkable plasticity of D. radiodurans criminated from the background fluorescence. This most likely nucleoids. Both single-molecule and ensemble microscopy explains why a reduced number of foci were counted per cell approaches were used to probe the dynamics of HU proteins in compared with the original study25.

NATURE COMMUNICATIONS | (2019) 10:3815 | https://doi.org/10.1038/s41467-019-11725-5 | www.nature.com/naturecommunications 7 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11725-5

abD distribution 200 0.15

160 0.1 ] 2 m μ

0.05 120 MSD [ Average MSD

0 Fitted lags Count 80 0 0.1 0.2 0.3 0.4 Time [s]

μ 2 –1 40 Diff. coeff. = 0.32 m ·s

2 μm 0 0 0.5 1 1.5 Diff. coeff. (μm2 ·s–1) c d 1.2

1.0

1 μm 0.8 0.6

0.4

Normalized fluo. Mobile fraction: 0.99

0.2 t1/2: 1.03 s

0.0 2468 Time (s)

Fig. 5 Nucleoid and HU dynamics in live D. radiodurans. a Reconstructed PALM image of HU-PAmCherry-expressing cells (stack of 15,000 frames with 50 ms frametime and acquired with constant 1 kW cm−2 561 nm laser and increasing 405 nm laser power). Several of the nucleoid morphologies illustrated in Fig. 3 can be seen in this image, but without a significative improvement in the perceived spatial resolution, due to the dynamics of nucleoids in live cells that change conformation in the second-to-minute timescale (Supplementary Movie 5). b Distribution of the apparent diffusion coefficients of 1355 single tracks of HU-PAmCherry measured by single-particle-tracking PALM in live, exponentially growing HU-PAmCherry-expressing D. radiodurans. Inset: mean square deviation (MSD) plot of the HU-PAmCherry tracks (Δt = 30 ms). The curve indicates that the molecules are confined with a confinement radius of 400 nm. The mean apparent diffusion coefficient derived from these 1355 tracks is of 0.32 µm2/s. c, d Ensemble measurements of HU-mCherry mobility probed by fluorescence recovery after photobleaching (FRAP) on HU-mCherry-expressing live D. radiodurans cells. c Examples of HU-mCherry-labelled nucleoids used for FRAP experiments. The photobleached region (ROI1) is indicated with a white box. Scale bar: 1 µm. d Analysis of the recovery of the fluorescence signal of HU-mCherry after photobleaching. Fitting this full-scale normalized data reveals that 99 ± 1% of HU-mCherry molecules are mobile exhibiting a half-life of 1.03 ± 0.06 s. Presented data are the mean values ± SD (N = 10). Source data are provided as a Source Data file

Throughout the cell cycle, the number of ter foci in cells was distribution and could be seen throughout the cell cytoplasm with systematically lower than that of oriC foci (Fig. 6a), as reported only a minor fraction localized along the new division axis and in previously25. The number of oriC foci was found to double as the the centre of the cells. Multiple oriC foci were thus seen to be cells progress from Phase 1 to Phase 5, going from a mean radially distributed around a single centrally located ter foci for number of 2.25 foci per cell to 4.44 foci per cell in Phase 5 most of the cell cycle. This was particularly visible in Phases 1 and (Fig. 6b). After septal closure, in Phase 6, the number of oriC foci 6. Interestingly, both ter and oriC seemed to be excluded from the was almost divided by 2, reaching 2.38 foci per cell. This suggests peripheral region of the cell bordering the cell wall. that DNA replication occurs all along the cell cycle until all copies No clear correlation could be established between foci localiza- of oriC from chromosome 1 have been duplicated. In contrast, the tion and nucleoid morphology. However, particular configurations number of ter foci remains much lower around 1.09 foci per cell could be seen (Fig. 6d). ter foci could, e.g., be observed occasionally and constant for most of the cell cycle until Phase 5 where it in the centre of the toroid- and square-shaped nucleoids, i.e., in increases to reach 1.38 foci per cell. regions with low DNA density, devoid of HU-mCherry labelling. In We also investigated the localization and distribution of the two the late stage of nucleoid segregation (Phase 5), ter foci were also loci within the cells (Fig. 6c). The ter foci were found to be seen to localize on the thin thread connecting the partially specifically located to the central region of cells during most of the segregated nucleoids (double rings), close to the new division axis cell cycle. It was only in Phases 1 and 2 that the ter foci were found and the closing septum. oriC, in contrast, was often seen at the to be slightly off-centre. As the cells progress from Phase 1 to 3, the periphery of the nucleoids and occasionally at the outer tips of the ter loci progressively move away from the S-1 septum to relocate in branched nucleoids observed in Phase 4 cells. the centre of the growing cells. In all phases, however, a majority of ter foci were located along the new division axis and 47% of them were still localized in this region at late stages of septal closure in Discussion Phase 5, just before cytokinesis, indicating a very late segregation In this work, we have characterized the cell cycles of exponentially of ter. In contrast, the oriC foci displayed a very different cellular and stationary growing D. radiodurans, enabling us to define

8 NATURE COMMUNICATIONS | (2019) 10:3815 | https://doi.org/10.1038/s41467-019-11725-5 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11725-5 ARTICLE

abter ori ter ori 2.5 6.0 2.0 5.0

1.5 4.0 3.0 HADA 1.0 Foci number Foci ParB-GFP 2.0 0.5 1.0 123456 123456 Phase Phase

c 0.35 ter localization density 0.30 + + + + + + 0.25

0.20

0.15 ori localization density 0.10 + + + + + + 0.05

0 Phase: 1 2 3 4 5 6

d HADA GFP mCherry Overlay

ter Toroid

ter Square

ter Double rings

ori Toroid

ori Square

ori Branched

Fig. 6 Distribution and choreography of ter and oriC loci of chromosome 1. a Images of D. radiodurans strains GY15787 and GY15800 expressing HU- mCherry and a heterologous ParB fused to GFP (green foci), with multiple copies of its cognate parS sequence inserted nearby chromosome 1 ter (left) or oriC (right) loci, respectively (Table 1), and stained with the cell wall HADA dye (blue). For clarity, the HU-mCherry signal is not included in these images. Scale bar: 2 µm. b Number of ter and oriC foci detected per cell at the different phases of the cell cycle retrieved from 3D images. Data are represented as mean ± SD. c Weighted distribution density of ter (top) and oriC (bottom) foci at the different phases of the cell cycle. The red cross indicates the centre of the cell. b, c ter (N = 315; N > 37 for each phase) or oriC (N = 368; N > 36 for each phase) foci were analysed and data were retrieved from two independent experiments. d Examples of ter and oriC foci localization relative to the different nucleoid conformations of D. radiodurans, expressing HU-mCherry (red) and ParB-GFP (green), and stained with the cell wall dye HADA (blue). Scale bar: 1 µm. Source data are provided as a Source Data file eight distinct phases, and have followed the changes in the cell observed: (i) between two cells that form a diad, resulting in cells and nucleoid morphology occurring during these processes. of the same diad being in different phases of the cell cycle, or (ii) We determined that exponentially growing D. radiodurans within a given cell, in which one side of the closing septum started diads grow throughout their cell cycle via remodelling of their growing earlier or faster than the other side, although, overall, no entire cell surface (peripheral and septal cell walls) and divide in difference in length was observed between the two sides of the alternate orthogonal planes. As proposed in an early study of D. closing septum (Supplementary Fig. 3). In exponentially growing 33 fi radiodurans , our 3D images con rmed that growth of the D. radiodurans, once the dividing S0 septum closes (Phase 6), the dividing septum occurs through a closing door mechanism and newly formed tetrads are relatively short-lived, lasting for a dozen not as a diaphragm as has been reported for ovococci41,42. One of minutes before splitting into two diads to initiate a new cell side of the S0 septum starts growing precisely from the middle of cycle (Phase 1). In Phase 6, cytokinesis results from the pro- the S-1 septum, whereas the other grows from the middle of the gressive increase in the curvature of the cell wall leading to a opposite peripheral cell wall, both at a 90° angle to the S-1 septum. reduction in the length of the shared S-1 septum, until the two This observation is surprising and will be the subject of future diads eventually separate, creating a small gap between the two studies. Septal growth lasts for two-thirds of the cell cycle (Phases adjacent diads (see separating tetrad, bottom left in Fig. 1a). It 3–6). Asynchronization of the division process was frequently was recently shown in S. aureus that the splitting of the diads into

NATURE COMMUNICATIONS | (2019) 10:3815 | https://doi.org/10.1038/s41467-019-11725-5 | www.nature.com/naturecommunications 9 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11725-5 single cells occurs on the millisecond timescale driven by the ‘sensed’ indirectly by chromosomes via the pressure exerted by turgor pressure35,43.InD. radiodurans, visualizing this splitting depletion forces induced by cytosolic crowders8,50, which would process was more difficult, as the cells do not undergo any sub- promote the compact shape and central localization of nucleoids stantial morphological changes upon cytokinesis. Instead, the in cells. In the case of D. radiodurans, several studies have separation appears to be slower and progressive, most likely described its nucleoid as being highly condensed26–30, but this catalysed by the enzymatic processing of the cell wall in this visual impression is in part due to the relatively large size of D. region of the cell. radiodurans cells that have a mean volume of ~ 2.6 μm3.We In every organism, nucleoid organization and cell division are therefore assessed the volume occupied by the nucleoids of D. intimately linked. The present study presents the first morpho- radiodurans and compared them with that of E. coli, which had logical characterization of the nucleoids of a coccus bacterium. previously been reported to occupy ~ 70% of the cell volume11. The nucleoids of D. radiodurans were found to display a Our measurements revealed that the mean volume of D. radio- remarkable plasticity, adopting multiple, distinct conformations durans nucleoids is very similar to that of E. coli (0.7 vs. 0.8 μm3); as the cells progress through their cell cycle. Our sptPALM and however, because of the larger cell volume, the fraction of the cell FRAP studies suggest that the dynamic nature of these nucleoids occupied by the nucleoid is significantly lower in the case of D. may be facilitated by the loose binding of the major D. radio- radiodurans and was found to vary from 0.30 to 0.35 depending durans NAP, the histone-like protein HU, to the genomic DNA. on the phase of the cell cycle. This fraction even decreased below To our knowledge, this is the first report of such a wide diversity 0.2 in stationary phase cells, which are smaller and present highly of nucleoid structures in bacteria. In addition to the previously condensed nucleoids. Such small cell fractions occupied by the reported toroid shape, the nucleoid was also found to adopt an nucleoid suggest either that depletion forces do not play a major elongated branched structure that aligned parallel to the S-1 role in nucleoid positioning and compaction in D. radiodurans or septum. This arrangement perpendicular to the newly growing S0 that D. radiodurans cells possess an unusually high concentration septum provides clear evidence that nucleoid organization and of crowders. chromosome segregation are tightly coupled to the cell division As D. radiodurans possesses multiple copies of its genome, process. In many model bacteria including S. aureus, E. coli and compared with a single copy in E. coli, the average DNA density B. subtilis, nucleoid occlusion (NO) systems, typically composed in D. radiodurans nucleoids is thus significantly higher than in E. of sequence-specific DNA-binding proteins, play a central role as coli, thereby confirming the impression that D. radiodurans spatial and temporal regulators of cell division by preventing the nucleoids are particularly condensed. Interestingly, the nucleoids assembly of the over the nucleoid44. Interestingly, in D. were found to be most compact in Phases 4 and 5, during the final radiodurans Phase 3 cells in which septal growth is starting, three stages of chromosome segregation. The specific positioning of the major conformations of nucleoids can be observed: toroids, nucleoids in the centre of cells and parallel to the S-1 septum, squares and crescent-shaped nucleoids formed by the opening of together with their increased levels of compaction during these the ring-like structures prior to their elongation and alignment late stages of the cell cycle, are reminiscent of metaphase chro- with the S-1 septum in Phase 4. Changes in the cell morphology mosomes in eukaryotic cells that align at mid-cell prior to seg- and, in particular, start of septal growth thus seem to precede the regation of each one of the two chromosomes into the two major rearrangements of the nucleoid, which suggests that the daughter cells. nucleoid is not the major determinant of the division site in D. Finally, we compared the number and distribution of the oriC radiodurans. Although no NO system has so far been identified in and ter loci of chromosome 1 as a function of the cell cycle. As D. radiodurans, we cannot exclude that such a system might also previously reported25, there were significantly more oriC foci in be at play in the regulation of cell division in this organism. cells than ter sites irrespective of the stage of the cell cycle. This The Min system that prevents the formation of the Z-ring in has been proposed to result from differences in the duration of aberrant locations45 might be the main factor ensuring that periods of transient cohesion of the oriC and ter loci from the cytokinesis occurs at mid-cell in D. radiodurans. Many spatial various copies of chromosome 1 after replication—this period features have been proposed as key determinants of pole-to-pole being longer for ter—and/or a very late replication of ter com- oscillations of Min in E. coli, such as the highly negative mem- pared with oriC25. Our cartography of the localizations of oriC brane curvature at the poles46, the longest axis of confinement47, and ter loci in D. radiodurans at each phase of the cell cycle also the longest possible distance for the diffusion of MinD48 or the revealed very distinct distribution patterns for the two loci and symmetry axes and scale of the cell shape49. D. radiodurans diads allow us to propose a first model for nucleoid organization and do not possess any poles per se. In Phase 1, cell ellipticity is close choreography in a coccus bacterium possessing a complex, mul- to 1 and cells possess an infinite number of symmetry planes (i.e., ticopy genome (Fig. 7). Chromosome arrangement in D. radio- every plane orthogonal to the S-1 septum). It is only when the durans is neither longitudinal nor transversal, but instead radial cells start growing in Phase 2 (with a major change in the length with oriC sites distributed all around centrally located ter sites. of the major axis and no change in minor axis length leading to a The ter loci were clearly retained and clustered by a yet unknown decrease in the ellipticity of the cells) that the number of sym- mechanism in the central region of the cells until just before metry planes is reduced to two, both orthogonal to the S-1 sep- cytokinesis (Fig. 7b, c). ter foci could indeed still be seen in the tum, one having its normal vector parallel to the long axis of the small gap formed by the closing septal doors in Phase 5 cells. cell and the other with its normal vector parallel to the second Segregation of replicated ter loci occurred at a very late stage of short axis. A possible oscillation of MinCDE along the long axis the cell cycle (Phase 5; Fig. 7c) and the duplicated ter loci then of the ellipsoidal-shaped cell, parallel to the S-1 septum, could be progressively migrated to the centre of the two new sister cells. In the basis for the precise placement and angle of growth of the contrast, oriC loci localized throughout the cell, with the exclu- dividing septum, orthogonal to the S-1 septum and along the sion from membrane proximal regions, ruling out any direct short axis of the cell. In such a scenario, the Min system would anchoring of oriC loci to the cell walls as has been observed in sense the geometry of the cell and start to oscillate in Phase 2. some bacteria51–53. Further work will be needed to understand As bacterial nucleoids fill a large fraction of the total cell the detailed choreography of this complex genome and how it volume, a common hypothesis suggests that the cell size and relates to the various nucleoid structures observed in this study, shape may play a major role in chromosome positioning and but our findings suggest that cell shape may critically influence sizing. It was recently proposed that cell shape and size were both nucleoid organization and chromosome arrangement.

10 NATURE COMMUNICATIONS | (2019) 10:3815 | https://doi.org/10.1038/s41467-019-11725-5 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11725-5 ARTICLE

abc oriC oriC P1 Chr I ter oriC ter oriC oriC

oriC oriC oriC oriC Chr I oriC Chr II P2 P1 ter

ter oriC ter oriC ter oriC

oriC oriC oriC

oriC oriC P2 Chr II oriC ter ter

Phase 1 Phase 5

Fig. 7 Radial arrangement and segregation of chromosome 1 in D. radiodurans. a D. radiodurans possesses multiple copies (4 to 10) of four replicons (2 chromosomes: Chromosome 1, Chr1, and chromosome 2, Chr 2, and two , P1 and P2), each of which possesses an (oriC) and a termination site (ter). For Chr1, the oriC and ter loci are represented in red and in green, respectively. b, c Proposed arrangement of the multiple copies of Chr1 in Phase 1 and 5 cells. Typical nucleoid shapes observed in these two phases (Fig. 3c) are shown in pale pink. Chr1 is present in at least four copies in Phase 1 and this number doubles by the end of the cell cycle (Phase 5). The number of ter foci remains constant and close to 1 for most of the cell cycle, most likely as a result of the clustering of ter loci by an unknown mechanism in the centre of the cell. The oriC loci are radially distributed around the ter loci and are found throughout the cytoplasm with the exception of membrane proximal regions

Methods Image acquisition. For PALM/PAINT imaging, wide-field illumination was Bacterial cultures. All D. radiodurans strains (wild-type, HU-mCherry GY15743, achieved by focusing the laser beams to the back focal plane of a 100 × 1.49 HU-PAmCherry GY17031 and oriC/ter-labelled strains GY15787 and GY15800; numerical aperture (NA) oil-immersion apochromatic objective lens (Olympus). Table 1) were grown at 30 °C with shaking in Tryptone-Glucos-Yeast extract 2x The intensity and time sequence of laser illumination at the sample were tuned by medium (TGY2x) supplemented with the appropriate . E. coli strain an acousto-optical tunable filter (AOTF; Quanta Tech). Near-circular polarization BL21 (DE3) (Table 1) was grown at 37 °C with shaking in Luria Bertani medium. of the laser beams was ensured by inserting a polychromatic quarter-wave plate For microscopy experiments, D. radiodurans cells were pre-grown the day before downstream the AOTF. Eight to 20 cells were imaged per field of view. Typically, fi and then diluted for an overnight growth until reaching exponential (OD650 ~0.3- for each sample, two or three elds of view on the same agarose pad were imaged 0.5) or stationary phase (OD650 > 5). To establish the growth curve of wild-type D. and at least three independent experiments were performed on different days. radiodurans cells, cultures inoculated at different cell densities were measured Sample drift was corrected in ImageJ using nanobeads (Sigma) deposited on every 2 h and then combined to obtain a complete growth curve. The final growth the agarose pads nearby the bacteria. curve corresponding to the mean of three independent experiments was fitted to an For single-particle-tracking experiments, data were acquired with continuous = −2 exponential growth curve (y y0*exp(K*x) with y0 corresponding to the y-value at 561 nm light illumination, at low power (130 W cm ) using a frametime of 30 ms. T = 0 and K the growth rate) using GraphPad Prism 6 and the doubling time was The Trackmate plugin for ImageJ was used to localize the particles in each frame calculated as ln2/K. and to connect the coordinates into trajectories57. Simple Linear Assignment Problem Tracker generated the tracks, with a maximal allowed linking distance set to 200 nm and a maximal frame interval between two spots to be bridged set to 2. Only tracks that contained more than four points were exported into MATLAB for Sample preparation for microscopy. Glass coverslides used for PALM/PAINT further processing. The TrackArt MATLAB software37 was used to analyse imaging of D. radiodurans were treated in an ozone oven for at least 10 min. Low- trajectories and to determine CPDs and apparent diffusion coefficients. Trajectories 54 melting agarose (1.75%; Biorad) pads dissolved in minimal medium (for PALM) shorter than eight frames or with a minimal individual mean square displacement or TGY2x (for spinning-disk microscopy) were poured on a cover slide inside a fit R2-values below 0.9 were filtered out. frame made with double-faced tape. A glass coverslip was placed on top of the Spinning-disk confocal microscopy was performed on the M4D cell imaging fl agarose pad, to atten the surface of the agarose while hardening. This coverslip platform at IBS using an Olympus IX81 microscope and Andor iXon Ultra fi was removed once the agarose had solidi ed. For imaging, bacterial cultures were EMCCD Camera with the laser beams focused to the back focal plane of a 100 × centrifuged 3 min at 3000 × g and were resuspended in 10 µl of medium or washed 1.49 NA oil-immersion apochromatic objective lens (Olympus). The intensity and in high-purity phosphate-buffered saline (GIBCO). One microlitre of this cell time sequence of laser illumination at the sample were tuned by an AOTF. For cell suspension was deposited on the pads. The drop was spread over the surface of the and nucleoid volume analyses, series of Z-planes were acquired every 100 nm using pad by rotating the coverslip. A second coverslip was placed over the agarose pad a piezo stage, whereas 200 nm was used for the multiplex imaging of oriC- and ter- containing the sample, thereby immobilizing the bacteria on the pad. labelled D. radiodurans strains. Fluorescence emission was collected through a To determine the growth phase and cell volumes, cells were incubated with Nile quad-band Semrock™ Di01-T405/488/568/647 dichroic mirror and the Red for 15 min at 30 °C, with agitation and then placed on an TGY agarose pad. corresponding single-band emission filters. For Nile Red time-lapse imaging For single time-point images acquired with the spinning-disk microscope, the Nile (Supplementary Movie 1), single-plane image sets (two-dimensional (2D)) were Red was used at a concentration of 30 µM. For correlative spinning-disk acquired every 10 min, for a total period of 4 h, using very low 561 nm laser power microscopy and PAINT imaging, the concentration was 300 nM. For PAINT (few micro-Watts) and 100 ms exposure times. Nucleoid time-lapse imaging of imaging alone, the concentration was 15 nM. To determine nucleoid volumes, cells Syto9- or HU-mCherry-labelled cells (Supplementary Movies 2 and 3) was in exponential phase were incubated with Nile Red (30 µM) and Syto9 (150 nM) for achieved by acquiring 3D images (Z-planes were acquired every 100 nm) every 5 15 min at 30 °C, with agitation and then placed on a TGY2x agarose pad. In all (Supplementary Movie 3) or 10 min (Supplementary Movie 2) for a period of 3 h. cases, the cells were rinsed to remove excess dye prior to imaging. An additional 3D time-lapse video of HU-mCherry was also performed with short For the PG labelling, cells in exponential growth phase were stained with the interval times (20 s) over a period of 10 min (Supplementary Movie 5). HU- fl 55 uorescent BADA dye at a concentration of 31 µM, incubated at 30 °C with mCherry time-lapses were corrected for bleaching using histogram matching in agitation for 15 min. Then Nile Red was added (15 µM) for another 15 min at 30 °C ImageJ58. with agitation. The cells were then rinsed in TGY2x at 30 °C, to remove unbound For FRAP measurements, a time-lapse acquisition of the fluorescence channel = dyes. Cells were then deposited on TGY2x agarose pads for imaging (T 0) or (mCherry) was performed for 60 timepoints at the fastest acquisition speed (10 = returned to the incubator until T 130 min. timepoints before and 50 timepoints after photobleaching). Photobleaching (Andor For time-lapse imaging of the cell growth and division process, cells in FRAPPA module) was set at the wavelength used for imaging, targeting a single exponential growth phase were incubated with Nile Red (15 µM) for 15 min at 30 ° point (crosshair) of the analysed nucleoid. Bleaching was performed using a low- C, with agitation. Cells were then rinsed in TGY2x before imaging. For time-lapse intensity setting (2% of the nominal maximum laser power), for 160 μs dwell time imaging of the nucleoids, exponentially growing HU-mCherry-expressing cells or and 12 repeats, to achieve a sufficiently small bleached region and with a loss of at Syto9-stained (150 nM for 5 min) wild-type cells were used. In all cases, cells were least 50% of the original signal intensity (mean signal loss of 60%). deposited on an TGY2x agarose pad designed with air holes to oxygenate the cells56. The samples were maintained at 30 °C during image acquisition. To image oriC-orter-labelled HU-mCherry-expressing D. radiodurans cells Data analysis. To calculate the volume of each cell, an ellipse was fitted to the (Table 1), an overnight pre-culture was diluted 60× in TGY2x supplemented with cellular membrane of Nile Red-labelled cells in 2D images (Supplementary Fig. 2a). −1 3.4 µg mL chloramphenicol and grown for a further 5 h until reaching an OD650 Measurements of the shorter (Lshort) and longer (Llong) axes of these ellipses were fi between 0.3 and 0.5. Cells were then stained with HADA (30 µM) for 30 min and extracted from such ts. The distance between the central S-1 septum and the deposited on a TGY2x agarose pad. extremity of the exterior of the ellipse along the short axis was measured (P), to

NATURE COMMUNICATIONS | (2019) 10:3815 | https://doi.org/10.1038/s41467-019-11725-5 | www.nature.com/naturecommunications 11 ARTICLE NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11725-5 retrieve the distance between the septum and the centre of the ellipse along the 8. Wu, F. et al. Cell boundary confinement sets the size and position of the E. coli short axis (w). With all these values, the total volume of the cell was computed chromosome. Curr. Biol. 29, 2131–2144 (2019). using the equations presented in Supplementary Fig. 2a with the cells approxi- 9. Esnault, E., Valens, M., Espéli, O. & Boccard, F. Chromosome structuring mated as prolate ellipsoids. The length and width of the elongated and toroid- limits genome plasticity in Escherichia coli. PLoS Genet. 3, e226 (2007). shaped nucleoids were measured as described in Supplementary Fig. 2c-d. 10. Dame, R. T. & Tark-Dame, M. Bacterial chromatin: converging views at To determine the extent of BADA (green) vs. Nile Red (red) staining in Phase different scales. Curr. Opin. cell Biol. 40,60–65 (2016). 1 cells, the perimeter of each cell was measured by fitting an ellipse to the Nile Red- 11. Fisher, J. K. et al. Four-dimensional imaging of E. coli nucleoid organization labelled cell membrane. The fraction of cell wall labelled with BADA was then and dynamics in living cells. Cell 153, 882–895 (2013). determined by measuring the two angles as shown in Supplementary Fig. 2b. The 12. Le, T. B., Imakaev, M. V., Mirny, L. A. & Laub, M. T. High-resolution perimeter of Nile Red-only-stained membranes (without any BADA labelling) for mapping of the spatial organization of a bacterial chromosome. Science 342, α an angle is retrieved from the equation (1). The BADA-labelled perimeter was 731–734 (2013). then determined by computing the lengths of the two sections of cell wall devoid of 13. Marbouty, M. et al. -and replication-mediated bacterial BADA labelling and subtracting them from the whole perimeter of the ellipse. rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi chromosome folding and origin condensation revealed by Hi-C and super- Zα  – 2 resolution imaging. Mol. Cell 59, 588 602 (2015). ¼ ðÞÃ ðÞθ 2þ Ã ðÞθ θ l Lshort cos Llong sin d ð1Þ 14. Umbarger, M. A. et al. The three-dimensional architecture of a bacterial 0 genome and its alteration by genetic perturbation. Mol. Cell 44, 252–264 To determine the volume of nucleoids, Z-stacks of the Syto9 fluorescence signal (2011). corresponding to the nucleoids were deconvoluted in Volocity using seven iterative 15. Wang, X., Liu, X., Possoz, C. & Sherratt, D. J. The two Escherichia coli – cycles and a calculated point spread function (PSF) (488 nm/520 nm). Cropped chromosome arms locate to separate cell halves. Dev. 20,1727 1731 (2006). images in which the fluorescence signal was homogeneous throughout the full field 16. Veiga, H., Jorge, A. M. & Pinho, M. G. Absence of nucleoid occlusion effector of view were then exported to the Imaris (Bitplane) software, where 3D objects Noc impairs formation of orthogonal FtsZ rings during Staphylococcus aureus corresponding to the nucleoids were created based on the automatic thresholding cell division. Mol. Microbiol. 80, 1366–1380 (2011). of the fluorescence signal. The volumes of these objects were then extracted for 17. Blasius, M., Hübscher, U. & Sommer, S. Deinococcus radiodurans: what further analysis. The same procedure and settings in Imaris were used to extract belongs to the survival kit? Crit. Rev. Biochem. Mol. Biol. 43, 221–238 (2008). nucleoid volumes from all our samples, including E. coli nucleoids. 18. Makarova, K. S. et al. Genome of the extremely radiation-resistant bacterium To evaluate the distribution of oriC and ter foci, Z-stacks of the GFP signal were Deinococcus radiodurans viewed from the perspective of comparative iteratively deconvoluted using a calculated PSF (488 nm/520 nm) in Volocity. In genomics. Microbiol. Mol. Biol. Rev. 65,44–79 (2001). the case of ter, Z-projections for each channel were concatenated into multi- 19. Slade, D. & Radman, M. resistance in Deinococcus channel stacks. In the case of oriC, the same multi-channel stacks were assembled, radiodurans. Microbiol. Mol. Biol. Rev. 75, 133–191 (2011). but we also verified that fluorescent foci were not lost by the Z-projection. 20. Daly, M. J. A new perspective on radiation resistance based on Deinococcus Fluorescent foci were manually marked in ImageJ to extract their xy coordinates as radiodurans. Nat. Rev. Microbiol. 7, 237–245 (2009). well as the dimensions and phase of the cells in which they were found. The 21. Tian, B., Xu, Z., Sun, Z., Lin, J. & Hua, Y. Evaluation of the antioxidant effects number of foci and their distribution were then automatically determined with of carotenoids from Deinococcus radiodurans through targeted mutagenesis, respect to the average cell dimension of D. radiodurans cells at each phase of the chemiluminescence, and DNA damage analyses. Biochim. et. Biophys. Acta cell cycle derived from PAINT imaging. 1770, 902–911 (2007). fl For FRAP data analysis, raw data corresponding to uorescence intensity 22. Wang, P. & Schellhorn, H. E. Induction of resistance to hydrogen peroxide measurements from two regions of interest (ROI1: bleached area, ROI2: total and radiation in Deinococcus radiodurans. Can. J. Microbiol. 41, 170–176 fl nucleoid) were corrected for background uorescence (ROI3: background) and (1995). 59 then normalized using the full-scale method in EasyFRAP , and the resulting 23. Battista, J. R. Against all odds: the survival strategies of Deinococcus fi 2 = = normalized data were tted (R 0.98) to the following equation y Mfrac*x/(t1/2 – + radiodurans. Annu. Rev. Microbiol. 51, 203 224 (1997). x) in GraphPad Prism 6 in which Mfrac corresponds to the mobile fraction and 24. Kitayama, S. & Matsuyama, A. Genome multiplicity and radiation resistance t1/2 to the half maximal recovery time of the mobile fraction. in radiodurans. J. Biochem. 90, 877–880 (1981). The non-parametric Mann–Whitney statistical tests were performed with 25. Passot, F. M. et al. Nucleoid organization in the radioresistant bacterium GraphPad Prism 6, to assess the difference in cell dimensions and nucleoid Deinococcus radiodurans. Mol. Microbiol. 97, 759–774 (2015). volumes during the cell cycle. Two-tailed P-values below 0.01 were considered as 26. Zimmerman, J. M. & Battista, J. R. A ring-like nucleoid is not necessary for significant and were indicated with asterisks as follows: **P ≤ 0.01, ***P ≤ 0.001, in the Deinococcaceae. BMC Microbiol. 5, 17 (2005). **** P ≤ 0.0001. 27. Englander, J. et al. DNA toroids: framework for DNA repair in Deinococcus radiodurans and in germinating bacterial . J. Bacteriol. 186, 5973–5977 Reporting summary. Further information on research design is available in (2004). the Nature Research Reporting Summary linked to this article. 28. Levin-Zaidman, S. et al. Ringlike structure of the Deinococcus radiodurans genome: a key to radioresistance? Science 299, 254–256 (2003). Data availability 29. Minsky, A., Shimoni, E. & Englander, J. Ring-like nucleoids and DNA repair through error-free nonhomologous end joining in Deinococcus radiodurans. J. The source data underlying Figs. 1d-e, 2d, 3c, 4a-b, 5b, 5d, 6b-c and Supplementary – Figs. 1d-e, 3a-c, 4b, 4d, 6, 7a, 8e and 9b-d are provided as a Source Data file. Other data Bacteriol. 188, 6047 6051 (2006). that support the findings of this study are available from the corresponding author upon 30. Minton, K. W. & Daly, M. J. A model for repair of radiation-induced DNA request. double-strand breaks in the extreme radiophile Deinococcus radiodurans. Bioessays 17, 457–464 (1995). 31. Toueille, M. et al. A comparative proteomic approach to better define Received: 11 March 2019 Accepted: 24 July 2019 Deinococcus nucleoid specificities. J. Proteom. 75, 2588–2600 (2012). 32. Daly, M. J. et al. Accumulation of Mn (II) in Deinococcus radiodurans facilitates gamma-radiation resistance. Science 306, 1025–1028 (2004). 33. Murray, R., Hall, M. & Thompson, B. Cell division in Deinococcus radiodurans References and a method for displaying septa. Can. J. Microbiol. 29, 1412–1423 (1983). 1. Luijsterburg, M. S., Noom, M. C., Wuite, G. J. & Dame, R. T. The architectural 34. Thornley, M. J., Horne, R. & Glauert, A. M. The fine structure of Micrococcus role of nucleoid-associated proteins in the organization of bacterial chromatin: radiodurans. Arch. Mikrobiol. 51, 267–289 (1965). a molecular perspective. J. Struct. Biol. 156, 262–272 (2006). 35. Monteiro, J. M. et al. Cell shape dynamics during the staphylococcal cell cycle. 2. Badrinarayanan, A., Le, T. B. & Laub, M. T. Bacterial chromosome Nat. Commun. 6, 8055 (2015). organization and segregation. Annu. Rev. Cell Dev. Biol. 31, 171–199 (2015). 36. Nguyen, H. H. et al. The essential histone-like protein HU plays a major role 3. Dillon, S. C. & Dorman, C. J. Bacterial nucleoid-associated proteins, nucleoid in Deinococcus radiodurans nucleoid compaction. Mol. Microbiol. 73, 240–252 structure and gene expression. Nat. Rev. Microbiol. 8, 185 (2010). (2009). 4. Wang, X., Llopis, P. M. & Rudner, D. Z. Organization and segregation of 37. Matysik, A. & Kraut, R. S. TrackArt: the user friendly interface for single bacterial chromosomes. Nat. Rev. Genet. 14, 191–203 (2013). molecule tracking data analysis and simulation applied to complex diffusion 5. Jun, S. Chromosome, cell cycle, and entropy. Biophys. J. 108, 785 (2015). in mica supported lipid bilayers. BMC Res. Notes 7, 274 (2014). 6. Pelletier, J. et al. Physical manipulation of the Escherichia coli chromosome 38. Kumar, M., Mommer, M. S. & Sourjik, V. Mobility of cytoplasmic, membrane, reveals its soft nature. Proc. Natl Acad. Sci. USA 109, E2649–E2656 (2012). and DNA-binding proteins in Escherichia coli. Biophys. J. 98, 552–559 (2010). 7. Shendruk, T. N., Bertrand, M., Haan, H. W., de, Harden, J. L. & Slater, G. W. 39. Gao, Y. et al. Charged residues in the H-NS linker drive DNA binding and Simulating the entropic collapse of coarse-grained chromosomes. Biophys. J. gene silencing in single cells. Proc. Natl Acad. Sci. USA 114, 12560–12565 108, 810–820 (2015). (2017).

12 NATURE COMMUNICATIONS | (2019) 10:3815 | https://doi.org/10.1038/s41467-019-11725-5 | www.nature.com/naturecommunications NATURE COMMUNICATIONS | https://doi.org/10.1038/s41467-019-11725-5 ARTICLE

40. Stracy, M. et al. Live-cell superresolution microscopy reveals the organization Acknowledgements of RNA polymerase in the bacterial nucleoid. Proc. Natl Acad. Sci. USA 112, We thank Professor Yves Brun and Professor Michael Van Nieuwenhze for providing us E4390–E4399 (2015). with the BADA dye. This work was supported by CEA Radiobiology grant (A-IRBIO-01- 41. Touhami, A., Jericho, M. H. & Beveridge, T. J. Atomic force microscopy of cell 19) and the CEA Irtélis programme (PhD grant of K.F.). This work used the M4D growth and division in Staphylococcus aureus. J. Bacteriol. 186,3286–3295 (2004). imaging platform of the Grenoble Instruct-ERIC Center (ISBG: UMS 3518 CNRS-CEA- 42. Wheeler, R., Mesnage, S., Boneca, I. G., Hobbs, J. K. & Foster, S. J. Super- UGA-EMBL) within the Grenoble Partnership for Structural Biology (PSB), supported resolution microscopy reveals cell wall dynamics and peptidoglycan by FRISBI (ANR-10-INBS-05-02) and GRAL (ANR-10-LABX-49-01), financed within architecture in ovococcal bacteria. Mol. Microbiol. 82, 1096–1109 (2011). the University Grenoble Alpes graduate school (Ecoles Universitaires de Recherche) 43. Zhou, X. et al. Mechanical crack propagation drives millisecond daughter cell CBH-EUR-GS (ANR-17-EURE-0003). IBS acknowledges integration into the Inter- separation in Staphylococcus aureus. Science 348, 574–578 (2015). disciplinary Research Institute of Grenoble (IRIG, CEA). 44. Wu, L. J. & Errington, J. Nucleoid occlusion and bacterial cell division. Nat. Rev. Microbiol. 10, 8 (2012). Author contributions 45. Lutkenhaus, J. Assembly dynamics of the bacterial MinCDE system and – K.F. and J.T. designed the research. K.F., F.L. and J.P.K. performed the microscopy spatial regulation of the Z ring. Annu. Rev. Biochem. 76, 539 562 (2007). fi 46. Renner, L. D. & Weibel, D. B. Cardiolipin microdomains localize to negatively experiments. P.S. provided the genetically modi ed strains of D. radiodurans. Y.S.W. curved regions of Escherichia coli membranes. Proc. Natl Acad. Sci. USA 108, synthesized the HADA dye. K.F., F.L., J.P.K., D.B. and J.T. analysed the data. K.F., D.B. 6264–6269 (2011). and J.T. wrote the manuscript and all authors discussed the results and approved the 47. Schweizer, J. et al. Geometry sensing by self-organized protein patterns. Proc. manuscript. Natl Acad. Sci. USA 109, 15283–15288 (2012). 48. Corbin, B. D., Yu, X.-C. & Margolin, W. Exploring intracellular space: Additional information function of the Min system in round-shaped Escherichia coli. EMBO J. 21, Supplementary Information accompanies this paper at https://doi.org/10.1038/s41467- 1998–2008 (2002). 019-11725-5. 49. Wu, F., Van Schie, B. G., Keymer, J. E. & Dekker, C. Symmetry and scale orient Min protein patterns in shaped bacterial sculptures. Nat. Nanotechnol. Competing interests: The authors declare no competing interests. 10, 719 (2015). 50. Jun, S. & Mulder, B. Entropy-driven spatial organization of highly confined Reprints and permission information is available online at http://npg.nature.com/ polymers: lessons for the bacterial chromosome. Proc. Natl Acad. Sci. USA reprintsandpermissions/ 103, 12388–12393 (2006). 51. Ben-Yehuda, S. et al. Defining a centromere-like element in by Peer review information: Nature Communications thanks John Battista and other identifying the binding sites for the chromosome-anchoring protein RacA. anonymous reviewer(s) for their contribution to the peer review of this work. Peer Mol. cell 17, 773–782 (2005). reviewer reports are available 52. Thanbichler, M. & Shapiro, L. MipZ, a spatial regulator coordinating chromosome segregation with cell division in Caulobacter. Cell 126, 147–162 Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in (2006). published maps and institutional affiliations. 53. Yamaichi, Y. et al. A multidomain hub anchors the chromosome segregation and chemotactic machinery to the bacterial pole. Genes Dev. 26, 2348–2360 (2012). Open Access This article is licensed under a Creative Commons 54. Floc’h, K. et al. Bacterial cell wall nanoimaging by autoblinking microscopy. Attribution 4.0 International License, which permits use, sharing, Sci. Rep. 8, 14038 (2018). 55. Pilhofer, M. et al. Discovery of chlamydial peptidoglycan reveals bacteria with adaptation, distribution and reproduction in any medium or format, as long as you give murein sacculi but without FtsZ. Nat. Commun. 4, 2856 (2013). appropriate credit to the original author(s) and the source, provide a link to the Creative 56. Jong, I. G. de, Beilharz, K., Kuipers, O. P. & Veening, J.-W. Live cell imaging of Commons license, and indicate if changes were made. The images or other third party ’ Bacillus subtilis and Streptococcus pneumoniae using automated time-lapse material in this article are included in the article s Creative Commons license, unless microscopy. J. Vis. Exp. 3145 (2011). indicated otherwise in a credit line to the material. If material is not included in the ’ 57. Tinevez, J.-Y. et al. TrackMate: an open and extensible platform for single- article s Creative Commons license and your intended use is not permitted by statutory particle tracking. Methods 115,80–90 (2017). regulation or exceeds the permitted use, you will need to obtain permission directly from 58. Schindelin, J. et al. Fiji: an open-source platform for biological-image analysis. the copyright holder. To view a copy of this license, visit http://creativecommons.org/ Nat. Methods 9, 676–682 (2012). licenses/by/4.0/. 59. Rapsomaniki, M. A. et al. easyFRAP: an interactive, easy-to-use tool for qualitative and quantitative analysis of FRAP data. Bioinformatics 28, © The Author(s) 2019 1800–1801 (2012).

NATURE COMMUNICATIONS | (2019) 10:3815 | https://doi.org/10.1038/s41467-019-11725-5 | www.nature.com/naturecommunications 13